How 'virtual rats' improve biological research studies

Researchers at the Medical College of Wisconsin don't need to make room in laboratory cages for their new test subjects. These rats are virtual -- actually computer models of rat physiology. I spoke last week with Daniel Beard, a computational biologist who developed the Virtual Rat Project. Below are excerpts from our interview.

On the benefits of 'virtual' rats:

The Virtual Rat Project [is not] an alternative to research on live rats. An important part of this program involves studying living animals in order to understand the physical parameters and estimate what we need to put into our computational models to predict function. You can't understand physiology simply by making measurements on animals. You need computer models to represent the complexity that's inherent in a biological system. The Virtual Rat Project helps us analyze the data we collect from an experiment and use the simulations to understand the underlying mechanisms. The next step is to determine the right experiment to do to learn something. From the perspective of using animals in research, the Virtual Rat Project lets us use animals much more efficiently and get a lot more information out of a given experiment.

[Virtual rat and live rat] studies go hand in hand. With simulating complex systems, we can use a model before we do any kind of experiment, to try to understand the state of the art. Let's put everything we know about a system into a simulation and see if it makes sense. Are the conclusions in the textbooks and scientific literature valid when you put them into a computer simulation? Until you put those things together, you're not going to know. As soon as you get beyond the trivial, you need a computer to make sense out of it. Rather than do more experiments, the model guides us on the key disconnections in the knowledge base. It allows us to generate some hypotheses that will potentially explain those disconnections or make those connections work. Then we have to do measurements to find out if it's right. Now we have a targeted, systematic way to go to the laboratory and measure what we need to measure.

If you're interested in drug action, the liver is very important. Having a virtual rat model allows us to take some information which comes from an in vitro study that's relevant to the liver. You can isolate enzymes from the liver and establish kinetic mechanisms for how a drug is metabolized. The Virtual Rat Project lets us integrate that information into a model of transporting the liver into whole-body transport of the drug. We can make a prediction about what that in vitro measurement means to the in vivo system. Then we can make a prediction about what in vivo measurements we need to find out how good or bad our prediction is. It's an iterative cycle.

On what the project looks like:

At this stage, the project is a set of computer models representing different aspects of physiology. Some are refined so they're at the level of simulating individual rat strains. In order to really make progress, we need to bring these models together with sources of data and make them work properly for the specific strains of rats we're studying and work together to simulate integrated function.

On the project's development and current phase:

We're not starting from scratch. There's a movement in the field toward not just toward integrative physiology using computer simulation, but also an increasing realization that to make progress we need large-scale, cooperative programs. Folks from around the world have been championing this idea for a long time. We need a lot of broad expertise, but also resources and tools to build modules or pieces of the puzzle that can talk to one another. This kind of coordinated effort has been going on for more than a decade. This is just one more program. It's well-funded. It's supported. This is a recognized effort.

We're always in the development phase. At the same time, this is not the kind of situation where we can start to do something five years from now once we're done. For example, we have a model of a kidney. That's very useful for analyzing and understanding data that relate to the things we can simulate in that model. We'll continue to make progress both with the kinds of scientific studies we can do with this particular model. At the same time, the ultimate goal is to integrate that kidney model into the whole cardiovascular function of the rat and then ultimately be able to understand the integrative function of the system. There are a lot of pieces in the puzzle. We can learn a lot, just by getting one of the pieces right. We're going to learn a great deal more by putting the puzzle together.

Photo: Daniel Beard

This post was originally published on Smartplanet.com